Research On Operation State Prediction And State Assessment Method Of HVDC Power Transmission Synchronous Condenser

With the rapid development of UHVDC transmission projects in China,as a powerful device for "strong reactive power support",the synchronous condenser can quickly provide large-scale dynamic reactive power to the power grid in a short period of time,so it has gradually been widely used in the power grid.However,the largecapacity synchronous condenser unit has a variety of equipment and complex structure,and the existing regular maintenance cannot meet the demand for safe operation of the large-capacity synchronous condenser.Therefore,realizing the prediction and anticipation of the operation status of the synchronous condenser system is of great significance to the safe operation of the converter station.In this paper,the following researches have been done on the prediction and comprehensive evaluation methods of the operation status of the synchronous condenser and its water and oil systems:Firstly,in view of the characteristics of non-linear,non-stationary and multisource interference factors in the monitoring signal of the synchronous condenser,a signal processing method based on feature engineering is proposed based on the monitoring data of the state monitoring system of the synchronous condenser.In this method,the global threshold denoising of wavelet transform is adopted to effectively eliminate the noise points,and the characteristic energy of vibration signals is extracted by combining the wavelet packet transform.The correlation degree between monitoring quantities is quantitatively analyzed by using the mutual information theory,so as to realize the optimal selection of characteristic indexes.Secondly,in order to solve the problems of Short operation time and less fault data of converter station,based on the vibration signal of the adjusting synchronous condenser bearing,a multivariable input matrix was constructed by fusing the other characteristic state variables.The Long-Short Term Memory Neural Network(LSTM)algorithm was introduced,and a multivariable state trend prediction model based on LSTM was proposed.According to the bearing vibration signal,combined with wavelet packet transform to effectively separate the low-frequency and high-frequency information of the signal,a vibration signal trend prediction model based on WPDLSTM is proposed to achieve the improvement of fitting effect and prediction accuracy.Through the accurate prediction of the variation trend of the state quantity,it provides a reference for the comprehensive evaluation of the health state of the system,and provides a method for data expansion.The validity and accuracy of the prediction method proposed in this paper are verified by comparing the prediction indexes between different models.Finally,combined with the LSTM prediction model,a health status evaluation model based on the dynamic deterioration degree of the synchronous condenser system is proposed.According to the correlation analysis result,the health status evaluation index system of the camera was constructed,and the variation trend of the evaluation indexes was included into the comprehensive evaluation model with the concept of dynamic deterioration degree.Finally,the comprehensive evaluation of the evaluation system was realized by studying the weight assignment method of combination weighting method and combining with the fuzzy theory.The research results of this paper are all based on the actual operation data of converter stations and the case of faults in the converter.The research ideas are clear,the research methods are simple and effective,conform to the actual situation of the project,and have certain engineering application value.